% latex table generated in R 4.3.1 by xtable 1.8-4 package
\begin{table}[ht]
\centering
\caption{Ethnic Minority Representation Effects on Turnout} 
\label{table:effect_turnout_w_covariates}
\begin{tabular}{lccc}
  \hline
 & (1) & (2) & (3) \\ 
  \hline
RD estimate & 0.043 & 0.077 & -0.003 \\ 
    & [0,0.095] & [0.007,0.165] & [-0.058,0.052] \\ 
  \% ethnic min & 0 & -0.002 & -0.001 \\ 
    & [-0.001,0.001] & [-0.004,0] & [-0.002,0.001] \\ 
  \% econ grade $>$=C & 0.354 & 0.65 & 0.068 \\ 
    & [-0.183,0.89] & [0.299,1.001] & [-0.809,0.946] \\ 
  \% low deprivation & -1.159 & -0.495 & -1.044 \\ 
    & [-1.791,-0.527] & [-1.551,0.56] & [-1.871,-0.216] \\ 
  pop density & 0 & -0.001 & 0 \\ 
    & [0,0.001] & [-0.003,0] & [0,0.001] \\ 
  incumbent & -0.003 & -0.004 & 0.017 \\ 
    & [-0.027,0.02] & [-0.032,0.023] & [-0.013,0.047] \\ 
  Labour & -0.03 & -0.019 & 0 \\ 
    & [-0.054,-0.006] & [-0.052,0.015] & [-0.029,0.029] \\ 
  Mean control & 0.673 & 0.64 & 0.696 \\ 
  R2 & 0.35 & 0.74 & 0.37 \\ 
  Num. eff. obs. & 106 & 44 & 51 \\ 
  Num. obs. & 465 & 258 & 207 \\ 
  MSE-opt bandwidth & 21.44 & 17.94 & 21.92 \\ 
   \hline
{\footnotesize Notes: The dependent variable is the turnout rate in general election \emph{t+1}. Average treatment effect at cutoff estimated with local linear regression with triangular kernel and MSE-optimal bandwidth. In brackets robust bias-corrected 95\% CI for the RD coefficient and conventional heteroskedasticity-consistent 95\% CI for the covariate coefficients.}
 \hline
\end{tabular}
\end{table}
